© 2008 American Thoracic Society doi: 10.1164/rccm.200808-1223ED
Etanercept for Idiopathic Pulmonary FibrosisLessons on Clinical Trial Design
University of Miami,
The search for an effective agent to treat idiopathic pulmonary fibrosis (IPF) continues. In this issue of the Journal (pp. 948–955) Raghu and colleagues report an important, albeit negative, result in a well-designed randomized controlled trial (RCT) of tumor necrosis factor (TNF)- The conclusions from the present RCT speak for themselves. The design raises interesting questions regarding the biology of TNF, the classical bleomycin model of pulmonary fibrosis, placebo controlled trials, endpoints that use surrogates for mortality, and classification of disease by phenotype in the postgenomic era.
The basic science underlying the hypothesis that TNF- The negative result of the etanercept study is unequivocal because the trialists used a placebo-controlled design, which is ethically acceptable when no known effective therapy exists. It would be less appropriate to conduct a trial in which both groups receive therapy of unknown efficacy (even though standard of care). Such a design makes results regarding the experimental intervention impossible to interpret; for example, it remains unclear whether N-acetylcysteine has a beneficial effect in patients with IPF because both groups received prednisone and azathioprine (9). The placebo-controlled design of the present trial is an important benchmark for future IPF studies.
Apart from mortality, the appropriate primary endpoints for IPF trials are arguable and undecided. The use of surrogate markers for mortality in clinical trials is attractive because it allows investigators to design trials of shorter duration, thus lowering costs and improving patient retention. There has been much interest in identifying physiologic variables that may be used as surrogates for mortality in IPF (10–12). This study used two measures of pulmonary function (FVC and diffusing capacity) and a measure of gas exchange, rather than mortality, as primary endpoints. Other IPF trials have used similar surrogate measures as primary outcomes (2–4). All have failed to show the efficacy of the treatment arm. These negative results may be because surrogate markers of mortality are insensitive to the effects of the experimental treatments. Another possibility is that the treatment outcomes used in murine model studies for IPF (prevention or resolution of scar tissue) vary greatly from those used in human trials (prevention of disease progression or decreased mortality). When investigators of the negative trial, who explored treatment of IPF with interferon-
The hazards of subgroup and retrospective analyses have been demonstrated to adversely affect patient care and business prospects. Authors of the present study assessed the effect of etanercept on death and disease progression ( The case definition used in this and other IPF clinical trials, based on the scheme proposed by the American Thoracic Society/European Respiratory Society in 2000 (14), is essentially a description of phenotype, requiring neither demonstration of usual interstitial pneumonia nor genetic analysis of known mutations that could lead to pulmonary fibrosis. Important recent work has demonstrated differences in gene expression between usual interstitial pneumonia and other interstitial pneumonias (15), genes that contribute to disease development in familial pulmonary fibrosis (16) and several potential candidate genes that may have a disease-modifying effect (16). As our knowledge of the genetic markers contributing to the development of IPF evolves, the current phenotypic definition should be refined to include specific genotypes. Hopefully, this knowledge will guide future trials testing specific molecular targets, such as etanercept, in the treatment of IPF. Patients and physicians will be disappointed by the negative results of this trial. However, the authors have established an important benchmark by demonstrating that a true placebo-controlled RCT is feasible and safe in IPF. An integrated exploration of the cellular pathways involved in the development of pulmonary fibrosis and the genes that regulate this process will provide us with new potential therapeutic targets and will further our understanding of the IPF phenotype. Until an effective therapy for IPF is found, placebo-controlled RCT designs should be used when investigating new drugs. Phase 3 trials of new drugs should use mortality as a primary outcome until surrogates, which are known to be adequate markers of mortality and which are affected by the pathway targeted by the investigational drug, are found. FOOTNOTES Conflict of Interest Statement: R.M.J. received $2,500 in 2005 for attendance at Intermune Advisory Board meetings. C.D.F. has no financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
Related articles in AJRCCM:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
antagonist therapy in patients with IPF (
1b, examined the efficacy of the study endpoints (all surrogate markers for mortality), they concluded that mortality was the best endpoint for therapeutic studies in IPF (
10% decrease in FVC). Although composite results in Raghu and colleagues' Figure 2 suggest that fewer such "events" occurred in the etanercept group, data in Table 6 show that mortality in etanercept subjects was in fact twice that of placebo-treated subjects. The investigators probably correctly assert that no deaths were due to the study drug. Predictable correlations were found among questionnaire responses and both FVC and six-minute-walk test data. Not surprisingly, patients whose FVC worsened experienced a greater decrease in quality of life.